Skip to content
GitLab
Explore
Sign in
Register
Primary navigation
Search or go to…
Project
dg
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Snippets
Build
Pipelines
Jobs
Pipeline schedules
Artifacts
Deploy
Releases
Package registry
Model registry
Operate
Environments
Terraform modules
Monitor
Incidents
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Terms and privacy
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
gmsh
dg
Commits
82e4a308
Commit
82e4a308
authored
3 years ago
by
Matteo Cicuttin
Browse files
Options
Downloads
Patches
Plain Diff
Slight RK4 Maxwell optimization, CPU.
parent
0dcb0e72
No related branches found
No related tags found
No related merge requests found
Changes
1
Show whitespace changes
Inline
Side-by-side
Showing
1 changed file
src/maxwell/maxwell_cpu.cpp
+126
-55
126 additions, 55 deletions
src/maxwell/maxwell_cpu.cpp
with
126 additions
and
55 deletions
src/maxwell/maxwell_cpu.cpp
+
126
−
55
View file @
82e4a308
...
@@ -238,6 +238,128 @@ compute_field_jumps(solver_state& state, const field& in)
...
@@ -238,6 +238,128 @@ compute_field_jumps(solver_state& state, const field& in)
}
}
}
}
/**************************************************************************
* Computational kernels: fluxes
*/
static
void
compute_field_jumps_and_fluxes
(
solver_state
&
state
,
const
field
&
in
)
{
double
jEx
=
0.0
;
double
jEy
=
0.0
;
double
jEz
=
0.0
;
double
jHx
=
0.0
;
double
jHy
=
0.0
;
double
jHz
=
0.0
;
for
(
auto
&
ed
:
state
.
eds
)
{
size_t
num_all_faces
=
num_elems_all_orientations
(
ed
)
*
ed
.
num_faces_per_elem
;
for
(
size_t
iF
=
0
;
iF
<
num_all_faces
;
iF
++
)
{
auto
n
=
ed
.
normals
.
row
(
iF
);
auto
nx
=
n
[
0
];
auto
ny
=
n
[
1
];
auto
nz
=
n
[
2
];
auto
face_det
=
ed
.
face_determinants
(
iF
);
for
(
size_t
i
=
0
;
i
<
ed
.
num_fluxes
;
i
++
)
{
auto
lofs
=
iF
*
ed
.
num_fluxes
+
i
;
auto
gofs
=
ed
.
flux_base
+
lofs
;
if
(
ed
.
fluxdofs_neigh
[
lofs
]
!=
NOT_PRESENT
)
{
auto
pos_mine
=
ed
.
fluxdofs_mine
[
lofs
];
auto
pos_neigh
=
ed
.
fluxdofs_neigh
[
lofs
];
jEx
=
in
.
Ex
[
pos_mine
]
-
in
.
Ex
[
pos_neigh
]
-
state
.
bndsrcs
.
Ex
[
gofs
];
jEy
=
in
.
Ey
[
pos_mine
]
-
in
.
Ey
[
pos_neigh
]
-
state
.
bndsrcs
.
Ey
[
gofs
];
jEz
=
in
.
Ez
[
pos_mine
]
-
in
.
Ez
[
pos_neigh
]
-
state
.
bndsrcs
.
Ez
[
gofs
];
jHx
=
in
.
Hx
[
pos_mine
]
-
in
.
Hx
[
pos_neigh
]
+
state
.
bndsrcs
.
Hx
[
gofs
];
jHy
=
in
.
Hy
[
pos_mine
]
-
in
.
Hy
[
pos_neigh
]
+
state
.
bndsrcs
.
Hy
[
gofs
];
jHz
=
in
.
Hz
[
pos_mine
]
-
in
.
Hz
[
pos_neigh
]
+
state
.
bndsrcs
.
Hx
[
gofs
];
}
else
{
auto
bc
=
state
.
matparams
.
bc_coeffs
[
gofs
];
auto
pos_mine
=
ed
.
fluxdofs_mine
[
lofs
];
jEx
=
bc
*
in
.
Ex
[
pos_mine
]
-
state
.
bndsrcs
.
Ex
[
gofs
];
jEy
=
bc
*
in
.
Ey
[
pos_mine
]
-
state
.
bndsrcs
.
Ey
[
gofs
];
jEz
=
bc
*
in
.
Ez
[
pos_mine
]
-
state
.
bndsrcs
.
Ez
[
gofs
];
jHx
=
(
2.0
-
bc
)
*
in
.
Hx
[
pos_mine
]
+
state
.
bndsrcs
.
Hx
[
gofs
];
jHy
=
(
2.0
-
bc
)
*
in
.
Hy
[
pos_mine
]
+
state
.
bndsrcs
.
Hy
[
gofs
];
jHz
=
(
2.0
-
bc
)
*
in
.
Hz
[
pos_mine
]
+
state
.
bndsrcs
.
Hz
[
gofs
];
}
auto
ndotE
=
nx
*
jEx
+
ny
*
jEy
+
nz
*
jEz
;
auto
ndotH
=
nx
*
jHx
+
ny
*
jHy
+
nz
*
jHz
;
auto
aE
=
face_det
*
state
.
matparams
.
aE
[
gofs
];
auto
bE
=
face_det
*
state
.
matparams
.
bE
[
gofs
];
auto
aH
=
face_det
*
state
.
matparams
.
aH
[
gofs
];
auto
bH
=
face_det
*
state
.
matparams
.
bH
[
gofs
];
/* Compute fluxes */
state
.
fluxes
.
Ex
[
gofs
]
=
aE
*
(
nz
*
jHy
-
ny
*
jHz
)
+
bE
*
(
ndotE
*
nx
-
jEx
);
state
.
fluxes
.
Ey
[
gofs
]
=
aE
*
(
nx
*
jHz
-
nz
*
jHx
)
+
bE
*
(
ndotE
*
ny
-
jEy
);
state
.
fluxes
.
Ez
[
gofs
]
=
aE
*
(
ny
*
jHx
-
nx
*
jHy
)
+
bE
*
(
ndotE
*
nz
-
jEz
);
state
.
fluxes
.
Hx
[
gofs
]
=
aH
*
(
ny
*
jEz
-
nz
*
jEy
)
+
bH
*
(
ndotH
*
nx
-
jHx
);
state
.
fluxes
.
Hy
[
gofs
]
=
aH
*
(
nz
*
jEx
-
nx
*
jEz
)
+
bH
*
(
ndotH
*
ny
-
jHy
);
state
.
fluxes
.
Hz
[
gofs
]
=
aH
*
(
nx
*
jEy
-
ny
*
jEx
)
+
bH
*
(
ndotH
*
nz
-
jHz
);
}
}
}
}
static
void
compute_fluxes_planar
(
solver_state
&
state
)
{
for
(
auto
&
ed
:
state
.
eds
)
{
size_t
num_all_faces
=
num_elems_all_orientations
(
ed
)
*
ed
.
num_faces_per_elem
;
for
(
size_t
iF
=
0
;
iF
<
num_all_faces
;
iF
++
)
{
auto
n
=
ed
.
normals
.
row
(
iF
);
auto
nx
=
n
[
0
];
auto
ny
=
n
[
1
];
auto
nz
=
n
[
2
];
auto
face_det
=
ed
.
face_determinants
(
iF
);
for
(
size_t
i
=
0
;
i
<
ed
.
num_fluxes
;
i
++
)
{
auto
lofs
=
iF
*
ed
.
num_fluxes
+
i
;
auto
gofs
=
ed
.
flux_base
+
lofs
;
/* Sources are imposed on jumps */
auto
jEx
=
state
.
jumps
.
Ex
[
gofs
]
-
state
.
bndsrcs
.
Ex
[
gofs
];
auto
jEy
=
state
.
jumps
.
Ey
[
gofs
]
-
state
.
bndsrcs
.
Ey
[
gofs
];
auto
jEz
=
state
.
jumps
.
Ez
[
gofs
]
-
state
.
bndsrcs
.
Ez
[
gofs
];
auto
jHx
=
state
.
jumps
.
Hx
[
gofs
]
+
state
.
bndsrcs
.
Hx
[
gofs
];
auto
jHy
=
state
.
jumps
.
Hy
[
gofs
]
+
state
.
bndsrcs
.
Hy
[
gofs
];
auto
jHz
=
state
.
jumps
.
Hz
[
gofs
]
+
state
.
bndsrcs
.
Hz
[
gofs
];
auto
ndotE
=
nx
*
jEx
+
ny
*
jEy
+
nz
*
jEz
;
auto
ndotH
=
nx
*
jHx
+
ny
*
jHy
+
nz
*
jHz
;
auto
aE
=
face_det
*
state
.
matparams
.
aE
[
gofs
];
auto
bE
=
face_det
*
state
.
matparams
.
bE
[
gofs
];
auto
aH
=
face_det
*
state
.
matparams
.
aH
[
gofs
];
auto
bH
=
face_det
*
state
.
matparams
.
bH
[
gofs
];
/* Compute fluxes */
state
.
fluxes
.
Ex
[
gofs
]
=
aE
*
(
nz
*
jHy
-
ny
*
jHz
)
+
bE
*
(
ndotE
*
nx
-
jEx
);
state
.
fluxes
.
Ey
[
gofs
]
=
aE
*
(
nx
*
jHz
-
nz
*
jHx
)
+
bE
*
(
ndotE
*
ny
-
jEy
);
state
.
fluxes
.
Ez
[
gofs
]
=
aE
*
(
ny
*
jHx
-
nx
*
jHy
)
+
bE
*
(
ndotE
*
nz
-
jEz
);
state
.
fluxes
.
Hx
[
gofs
]
=
aH
*
(
ny
*
jEz
-
nz
*
jEy
)
+
bH
*
(
ndotH
*
nx
-
jHx
);
state
.
fluxes
.
Hy
[
gofs
]
=
aH
*
(
nz
*
jEx
-
nx
*
jEz
)
+
bH
*
(
ndotH
*
ny
-
jHy
);
state
.
fluxes
.
Hz
[
gofs
]
=
aH
*
(
nx
*
jEy
-
ny
*
jEx
)
+
bH
*
(
ndotH
*
nz
-
jHz
);
}
}
}
}
#ifdef USE_MPI
#ifdef USE_MPI
static
void
static
void
exchange_boundary_data
(
const
model
&
mod
,
solver_state
&
state
)
exchange_boundary_data
(
const
model
&
mod
,
solver_state
&
state
)
...
@@ -458,57 +580,6 @@ exchange_boundary_data_H(const model& mod, solver_state& state)
...
@@ -458,57 +580,6 @@ exchange_boundary_data_H(const model& mod, solver_state& state)
}
}
#endif
/* USE_MPI */
#endif
/* USE_MPI */
/**************************************************************************
* Computational kernels: fluxes
*/
static
void
compute_fluxes_planar
(
solver_state
&
state
)
{
for
(
auto
&
ed
:
state
.
eds
)
{
size_t
num_all_faces
=
num_elems_all_orientations
(
ed
)
*
ed
.
num_faces_per_elem
;
for
(
size_t
iF
=
0
;
iF
<
num_all_faces
;
iF
++
)
{
auto
n
=
ed
.
normals
.
row
(
iF
);
auto
nx
=
n
[
0
];
auto
ny
=
n
[
1
];
auto
nz
=
n
[
2
];
auto
face_det
=
ed
.
face_determinants
(
iF
);
for
(
size_t
i
=
0
;
i
<
ed
.
num_fluxes
;
i
++
)
{
auto
lofs
=
iF
*
ed
.
num_fluxes
+
i
;
auto
gofs
=
ed
.
flux_base
+
lofs
;
/* Sources are imposed on jumps */
auto
jEx
=
state
.
jumps
.
Ex
[
gofs
]
-
state
.
bndsrcs
.
Ex
[
gofs
];
auto
jEy
=
state
.
jumps
.
Ey
[
gofs
]
-
state
.
bndsrcs
.
Ey
[
gofs
];
auto
jEz
=
state
.
jumps
.
Ez
[
gofs
]
-
state
.
bndsrcs
.
Ez
[
gofs
];
auto
jHx
=
state
.
jumps
.
Hx
[
gofs
]
+
state
.
bndsrcs
.
Hx
[
gofs
];
auto
jHy
=
state
.
jumps
.
Hy
[
gofs
]
+
state
.
bndsrcs
.
Hy
[
gofs
];
auto
jHz
=
state
.
jumps
.
Hz
[
gofs
]
+
state
.
bndsrcs
.
Hz
[
gofs
];
auto
ndotE
=
nx
*
jEx
+
ny
*
jEy
+
nz
*
jEz
;
auto
ndotH
=
nx
*
jHx
+
ny
*
jHy
+
nz
*
jHz
;
auto
aE
=
face_det
*
state
.
matparams
.
aE
[
gofs
];
auto
bE
=
face_det
*
state
.
matparams
.
bE
[
gofs
];
auto
aH
=
face_det
*
state
.
matparams
.
aH
[
gofs
];
auto
bH
=
face_det
*
state
.
matparams
.
bH
[
gofs
];
/* Compute fluxes */
state
.
fluxes
.
Ex
[
gofs
]
=
aE
*
(
nz
*
jHy
-
ny
*
jHz
)
+
bE
*
(
ndotE
*
nx
-
jEx
);
state
.
fluxes
.
Ey
[
gofs
]
=
aE
*
(
nx
*
jHz
-
nz
*
jHx
)
+
bE
*
(
ndotE
*
ny
-
jEy
);
state
.
fluxes
.
Ez
[
gofs
]
=
aE
*
(
ny
*
jHx
-
nx
*
jHy
)
+
bE
*
(
ndotE
*
nz
-
jEz
);
state
.
fluxes
.
Hx
[
gofs
]
=
aH
*
(
ny
*
jEz
-
nz
*
jEy
)
+
bH
*
(
ndotH
*
nx
-
jHx
);
state
.
fluxes
.
Hy
[
gofs
]
=
aH
*
(
nz
*
jEx
-
nx
*
jEz
)
+
bH
*
(
ndotH
*
ny
-
jHy
);
state
.
fluxes
.
Hz
[
gofs
]
=
aH
*
(
nx
*
jEy
-
ny
*
jEx
)
+
bH
*
(
ndotH
*
nz
-
jHz
);
}
}
}
}
static
void
static
void
compute_fluxes_planar_E
(
solver_state
&
state
)
compute_fluxes_planar_E
(
solver_state
&
state
)
{
{
...
@@ -599,13 +670,13 @@ compute_fluxes_planar_H(solver_state& state)
...
@@ -599,13 +670,13 @@ compute_fluxes_planar_H(solver_state& state)
static
void
static
void
compute_fluxes
(
const
model
&
mod
,
solver_state
&
state
,
const
field
&
in
,
field
&
out
)
compute_fluxes
(
const
model
&
mod
,
solver_state
&
state
,
const
field
&
in
,
field
&
out
)
{
{
compute_field_jumps
(
state
,
in
);
#ifdef USE_MPI
#ifdef USE_MPI
compute_field_jumps
(
state
,
in
);
exchange_boundary_data
(
mod
,
state
);
exchange_boundary_data
(
mod
,
state
);
#endif
/* USE_MPI */
compute_fluxes_planar
(
state
);
compute_fluxes_planar
(
state
);
#else
compute_field_jumps_and_fluxes
(
state
,
in
);
#endif
/* USE_MPI */
for
(
auto
&
ed
:
state
.
eds
)
for
(
auto
&
ed
:
state
.
eds
)
{
{
...
...
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment